In recognition of the importance of patient values in medical decisions, researchers have begun developing decision aids that inform patients about the probability of various harms and benefits associated with available treatment alternatives. Decision aids have great potential to help patients make informed decisions. But they could also bias patients' decisions. Our overall objective is to identify cognitive biases that are likely to occur when patients use decision aids and test ways to reduce these biases. We have identified four cognitive biases that are especially relevant to the design of decision aids: Framing effects: People's treatment choices depend on whether treatment benefits are framed in terms of survival or mortality. Undue influence of patient testimonials: Patient testimonials regarding the desirability or undesirability of a treatment influence treatment choices independent of statistical information about treatment effectiveness. Preference reversal after addition of a third treatment option: A person who prefers treatment option A over B may prefer B after the addition of a third option. Undue influence of outcome information: The addition of information about treatment outcomes may unduly influence patients' choices. We will develop decision aid materials using written and computerized surveys administered to the general public, with one-third of the sample having cancer-related experience. Our goal is not to test decision aids on patients facing difficult clinical situations but, instead, to develop methods by which such decision aids can be produced - general principles for the creation of decision aids that will increase patients' understanding of their treatment choices without introducing cognitive bias.